The term "neighborhood" is widely used in public health research, and throughout this chapter, in part because it is a component of the popular contemporary lexicon. However, it is a concept that is difficult to precisely define and measure and many definitions exist across various disciplines. For example, in the sociological tradition, a neighborhood is an ecologic structure nested within a local community and includes a collection of people and institutions occupying a spatially defined area (Park, 1915; Suttles, 1972). This area is influenced by both internal and external forces such as historical, cultural, ecological, environmental, geographical, and political forces (Park, 1915; Suttles, 1972). Based on traditional definitions, neighborhoods are often assumed to be nested and bounded structures that are discrete and non-overlapping (Chaskin, 1997). Additionally it is often assumed that neighborhoods need to be relatively homogeneous with respect to the exposures of interest (Pickett and Pearl, 2001). We will discuss alternative conceptualizations of neighborhoods that rely on a more "relational" as opposed to "spatial" understanding of neighborhoods (see Section 3.1.1).
In practice, neighborhoods are most often operationalized using administrative boundaries constructed by administrative agencies. In the US context, these boundaries are most consistently provided by the US Census Bureau and include counties, tracts, and block groups that vary in size between an average of 100,000 individuals per area (counties) to 1000 individuals per area (block groups) (U.S. Bureau of the Census, 1994). Zip codes, derived from the US Postal Service to facilitate efficient mail delivery, are also frequently used in the US context (U.S. Postal Service, 2003). There are a number of limitations to the use of administrative boundaries and a continuing debate over which of these areas is most appropriate and if they should be used at all (Krieger et al,
2002). Administrative boundaries are imperfect and considered mere proxies for the true geographical areas based on resident perceptions or historical and local knowledge. Additionally, different administrative boundaries vary dramatically in spatial scale. For example, census counties and zip codes are generally considered too large especially in trying to define homogeneous neighborhood exposures, a fundamental assumption of the statistical techniques used to create neighborhood variables and to relate them to health. Specifically, these methods require exposures that are more homogeneous than not in order to isolate neighborhood effects from the effects of compositional (individual-level) characteristics (Diez-Roux, 2000; Subramanian et al,
2003). In this regard, census tracts and block groups are preferred because they are smaller and were designed to be relatively homogeneous with respect to census-derived population and socioeconomic indicators. However, physical changes (i.e., changes in street patterns, highway construction, and other developments) as well as population growth and migration patterns cause the composition and boundaries of tracts and block groups to change over time. Moreover, because these areas are defined based on having an average number of individuals located within them, less densely populated areas are much larger in spatial scale as compared with more densely populated areas. For example, based on the 2000 US Census, tracts in more densely populated urban areas such as New York City, New York, are as small as 0.01 square miles. However, tracts in more rural and sparsely populated areas such as Santa Barbara, California, are as large as 1168 square miles (U.S. Bureau of the Census, 2000).
There are a number of benefits to using administratively defined neighborhood boundaries, which explains their wide use in the literature. Administrative data are readily available at no additional cost to researchers providing easy links to health outcomes data. In addition to pre-specified neighborhood boundaries, censuses also provide information on the sociode-mographic characteristics of residents and these aggregate socioeconomic status (SES) measures are often used as proxies for neighborhood features in analyses. Measures based on administrative areas can also be obtained for very large geographic regions allowing for national analyses. Despite the fact that neighborhoods based on administrative definitions are likely to be very imperfect proxies for the spatial areas relevant to health, the fact that they have allowed detection of health-relevant associations suggests that features of these areas may be correlated with the true area-level constructs of interest.
Researchers have also attempted to characterize neighborhoods based on historical roots or resident perceptions. For example, mental maps have been developed as a technique of creating neighborhood boundaries based on residents' perceptions (Downs and Stea, 1973; Gould and White, 1974). This approach requires individuals to construct cognitive maps representing their relationship to space including the geographic boundaries that defined their daily lives based on physical (i.e., streets, highways, rivers, landmarks) and social elements (i.e., relationship and networks of friends and neighbors). These maps remain underused in the public health literature because it is hard to conduct them on a large scale and because there is a great deal of variability in resident perceptions. Studies have documented that perceptions of neighborhood boundaries vary significantly by individual characteristics such as age, gender, race/ethnicity, and social class (Anderson, 1990; Chaskin, 1997). Moreover, this approach is often dismissed as being overly subjective and impractical because of the difficulty in obtaining objective measures for these subjectively defined areas.
A more fundamental issue related to defining and operationalizing neighborhood boundaries is identifying the appropriate spatial scale. The different spatial scales represented in the literature are in part driven by data availability such that the most common option is the use of administrative sources with pre-specified boundaries at various spatial scales. However, the most relevant spatial scale for investigation depends on many other factors including: (1) the processes through which area features are hypothesized to affect specific health outcome, (2) the neighborhood exposures being measured, and (3) the most appropriate spatial scale for intervention or policy-relevant solutions. Different spatial scales may be more or less relevant for specific health processes under investigation. For example, the immediate area (i.e., smaller neighborhood boundaries) may be important for understanding how environmental exposures (i.e., toxins) relate to asthma. Alternatively, larger areas may be more appropriate for understanding how the presence of fast food restaurants shapes dietary behaviors. Misspecifying the spatial scale can result in bias consistent with exposure misclassification in epi-demiologic research. Specifically, neighborhood exposures may be misclassified if the spatial scale most relevant to the process of interest is not used. This may result in spurious associations or lack of associations between neighborhood exposures and health outcomes. This problem is often referred to as the modifiable area unit problem (MAUP), the fact that the association detected between a geographic unit and an outcome is a function of the spatial scale used (Openshaw, 1984). Thus, using the wrong spatial scale or the wrong boundary within a correct spatial scale can produce biased results.
Because public health researchers are interested not only in studying problems but also in implementing solutions, decisions regarding the relevant spatial scale may be based on the potential for interventions or policy solutions. In some states (with more populated counties), local county governance determines how important resources such as courts, judicial systems, public transportation, welfare services, child and family services, hospitals, food and safety regulations, and environmental health services are distributed. As such, examining variations in health by counties may lead to results that are more amenable to local policy solutions. Similarly, considering definitions based on urban planning may also be useful. For example the city of New York consists of five boroughs and 59 community districts (New York City Department of City Planning, 2005). There are also hundreds of neighborhoods within these community districts with strong historical underpinning. However, goods and services are distributed at the community district level. Specifically, the department of city planning promotes strategic growth and development within each of the community districts providing an added incentive for investigating how features of these areas may relate to health.
Future research may benefit from a more careful consideration of the relevant spatial scale. However, there is limited theory available to inform our understanding on the appropriate spatial scale. For some research questions, multiple spatial scales may operate to affect the health processes under investigation. Additionally, although neighborhoods are most often studied in relation to health, there may be features of larger geographic areas such as metropolitan areas, states, and regions that have important health implications. Thus, it is important to understand how neighborhoods are located within surrounding areas and how features of these non-residential or surrounding environments may also matter for health. To this end, a growing body of research has examined neighborhood processes in relation to multiple spatial scales and has considered both local neighborhoods and surrounding neighborhoods in relation to health (Auchincloss et al, 2007; Chaix et al, 2005; Morenoff, 2003; Robert and Ruel, 2006). For example, Auchincloss and colleagues (2007) investigated associations between neighborhood poverty and insulin resistance, considering both neighborhood poverty within the local neighborhood and the distance to a wealthy neighborhood. They found that although neighborhood poverty was positively associated with insulin resistance, the distance to a wealthy area was also positively associated with insulin resistance. These findings suggest that while living in a poor neighborhood is problematic, so is living in close proximity to other poor neighborhoods.
In a similar vein, researchers are also beginning to conceptualize and operationalize neighborhoods based on more relational considerations. For example, some researchers argue that contrary to popular belief, neighborhoods are actually unstructured and unbounded constellations of connections and interactions between people (Castree, 2004; Graham and Healey, 1999). This conceptualization highlights the fluid and dynamic nature of neighborhood environments and challenges assumptions regarding our ability to tease out contextual versus compositional effects. The reality is that neighborhoods shape people and people shape neighborhoods and these relationships change over time (Cummins et al, 2007).
As a final note, neighborhoods and other geographic areas represent only one of many important contexts that may have health-relevant properties. Other important contexts such as school and work environments have also been studied in relation to health. Future research may benefit from considering how individuals exist within multiple overlapping contexts which may matter for health (Muntaner et al, 2006; Szapocznik et al, 2006). Statistical methods such as cross-classified random effects models make this type of investigation possible (Subramanian et al, 2003).
Was this article helpful?
Diabetes is a disease that affects the way your body uses food. Normally, your body converts sugars, starches and other foods into a form of sugar called glucose. Your body uses glucose for fuel. The cells receive the glucose through the bloodstream. They then use insulin a hormone made by the pancreas to absorb the glucose, convert it into energy, and either use it or store it for later use. Learn more...